The smell of damp limestone and old mortar is something you never really shake once it gets into your clothes. It’s a heavy, mineral scent, the kind that reminds you that gravity eventually wins every argument if you wait long enough. I spent most of my twenties working with a trowel, resetting the bones of buildings that had been standing since before the city had a paved road.

In masonry, you learn very quickly that a structure doesn’t usually collapse because of a single catastrophic blow. It fails because of the “reasonable” cracks-the small, weathered gaps that everyone decides are too minor to fix right now.

Last Tuesday, at 10:14 AM, I sat in a sleek boardroom in downtown Toronto. The air there didn’t smell like limestone; it smelled like expensive roasted coffee and the faint, ozone tang of high-end ventilation. On the far wall, a projector cast a crimson number onto the screen: 34.

This was the mobile performance score for a client’s homepage. In the room, fourteen people sat around a table that likely cost more than my first truck.

A red 34 is a failure. It is the digital equivalent of a foundation that has shifted three inches to the left. The developer at the end of the table, a woman named Sarah who looked like she hadn’t slept since the previous fiscal quarter, clicked to the next slide. This slide was a list. It contained fourteen lines of code-scripts, pixels, and snippets-that lived in the header of the website.

The Inventory of Bloat

There was the primary analytics suite, a second “backup” analytics tool that the CFO preferred, a session recording script to watch mouse movements, and a chat widget that popped up like an overeager greeter. Then came the two different attribution pixels for social media platforms, a consent banner required by legal, a scheduling embed for the sales team, a review widget for social proof, and a massive marketing automation script that handled everything from email captures to lead scoring.

Eleven of these scripts had a name written next to them in parentheses. These were the names of the people sitting in the room. Each person was the “owner” of that specific tool. Each tool represented a line item in their specific departmental budget. Each one had been added because of a single, perfectly logical sentence uttered months or years ago.

“We need to see where the leads are coming from,” said the head of growth.

“The sales team needs that calendar on the front page,” said the VP of Sales.

“We can’t risk losing data during the migration,” said the data analyst.

The conversation that followed lasted exactly four minutes. It was polite. It was professional. It was utterly useless. No one suggested removing their own script. Instead, the consensus shifted toward things that didn’t have a human advocate in the room.

“Maybe we should look at the image sizes,” someone suggested. “Or maybe the animations are too heavy,” another added. They looked at Sarah. She stared at the red 34 and nodded, knowing she had no power to touch the fourteen things actually causing the rot.

Governance is a word that people usually associate with boring board meetings and thick binders of rules. In reality, performance on the web is a governance property. We like to think of speed as a technical attribute, something that a skilled engineer can simply “build in.” But you cannot build a fast house on a foundation that everyone is allowed to chip away at whenever they feel like adding a new decorative tile.

Optimizing the Mortar vs. Fixing the Structure

When a site takes six seconds to load on a mid-range phone, it isn’t because the code is bad. It’s because the organization is incapable of saying “no” to a tool that promises a 2% lift in tracking accuracy at the cost of a 20% drop in user experience.

To fix the chimney, I had to take the stones off. There was no way to make the bricks “stronger” through sheer force of will. On the web, we try to solve the “heavy chimney” problem by asking developers to optimize the mortar. We ask for better compression, for lazy loading, for more efficient execution of the very scripts that shouldn’t be there in the first place.

We treat the symptom because the cure-deleting a script owned by a senior stakeholder-is politically expensive.

Most people don’t realize that a single marketing script can easily weigh more than the entire rest of the website’s visual assets combined. A single tracking pixel might trigger a chain reaction of four other requests, dragging in megabytes of JavaScript that has to be parsed and executed before a user can even see a headline.

If you have fourteen of these, you aren’t running a website; you’re running a middleman for data brokers that happens to have a logo at the top. The irony of the situation is that the very tools meant to “optimize” the business are the ones killing the conversion rate.

You cannot record a user’s session if the user leaves before the page loads. You cannot attribute a sale that never happens because the customer got frustrated and clicked away.

This is where the specialized approach of a studio like

Coherent Agency

becomes a different kind of value proposition. It’s not just about knowing how to write GSAP animations or how to structure a Webflow CMS. It’s about having a partner who can sit in that room and point at the red 34 with the authority of someone who knows that technical debt is actually organizational debt.

When we build things, we often forget that the most important part of a structure isn’t what you see-it’s what it can support. A website is a bridge between a brand and a human being. Every script you add is a toll booth on that bridge. Eventually, if you add enough toll booths, people just stop crossing the river.

I remember a specific job on a heritage home where the owner wanted to add a massive stone gargoyle to a porch that was already sagging. He had a great reason for it: his grandfather had owned one just like it. I told him I wouldn’t do it. Not because I didn’t like the gargoyle, but because I didn’t want to be the one who built the thing that eventually crushed his front door.

In the digital world, we rarely have the luxury of saying no. The “scripts” just keep coming. They arrive in Slack messages and “urgent” emails. They are the artifacts of a culture that values the collection of data over the experience of the person providing it.

The Tragedy of the Performance Commons

We treat the browser like a bottomless bucket, forgetting that on the other end of that connection is a person with a limited data plan and a diminishing amount of patience. The problem isn’t the analytics tool itself. It’s the fact that the analytics tool is protected by a department, while the speed of the site is protected by nobody.

Performance is a common good, and like all common goods, it is easily exploited by individuals pursuing their own narrow goals. The head of marketing gets their data. The sales lead gets their widget. The user gets a blank white screen and a spinning loading icon.

Editing is a violent act. It requires taking things that are perfectly good and throwing them away for the sake of the whole. It requires looking at a script that “only takes 200ms to load” and realizing that 200ms, multiplied by fourteen, is the difference between a sale and a bounce.

If you look at the source code of most modern B2B sites, you can see the history of the company’s internal politics written in the header tags. You can see the ghost of the marketing director who left three years ago but whose favorite heatmap tool is still running. You can see the remnants of the failed A/B test from last summer that no one bothered to turn off.

It’s a digital landfill, and the developers are the ones tasked with making the trash smell like roses. The masons who built the cathedrals of Europe didn’t have to worry about marketing scripts, but they understood the principle of the “reasonable” addition.

They knew that if you keep adding weight to the top, you eventually have to widen the base. On the web, we don’t have a wider base. We have the same mobile processor and the same 5G connection. The only thing that can change is the load.

Firing the Tools

To fix a slow site, you don’t need a faster server. You need a leader who is willing to walk into a room, look at fourteen people, and tell thirteen of them that their favorite tool is fired.

You need someone who values the silence of a fast-loading page more than the noise of a crowded dashboard. Until then, the number will stay red, the foundation will keep shifting, and we’ll all keep sitting in meetings talking about image sizes while the building slowly sinks into the dirt.